Auction Catalog Pages Built For A Phone And A Weak Signal

An auction catalog built for a phone and a weak signal is one where the first screen of lots arrives in about a second on mobile data, every photo below it waits until it is needed, and a bidder can jump to lot 214 without scrolling past two hundred others. That is the whole design brief, and it is a harsher one than most web work, because the place a bidder actually opens your catalog is a barn, a truck cab, a warehouse aisle or a preview hall with two bars of signal and a lot of concrete.

Catalogs that look fine on a desk in a browser tab fail in those places constantly. This article covers what to do about it: photo weight and formats, what loads and when, how to group and filter lots so the structure matches the sale rather than the database, when a lot deserves its own page, where condition reports and provenance belong, how to write a lot title someone actually searches for, the pagination question, and the printed catalogue that still matters in the hall.

Where bidders actually open your catalog

Bidders open the catalog in three places, and none of them is a comfortable one. The first is the night before the sale, on a phone, in bed, going through lots quickly to decide whether the drive is worth it. The second is at the preview, standing in front of a lot with the catalog open in one hand, checking the description against the thing itself. The third is during the sale, on a live simulcast or a timed sale, with a closing clock running and no patience at all for a page that takes four seconds to show a photo.

All three share the same constraints. One hand. A small screen. Bright light, often outdoors. And connectivity that ranges from decent to nearly nothing, because farm and heavy equipment sales happen on farms, estate sales happen in houses with thick walls, and industrial liquidations happen inside steel buildings that block signal well.

This is why generic catalog advice built for online retail does not transfer cleanly. A retail shopper who has a slow page can come back tomorrow. A bidder watching a lot close in ninety seconds cannot. The performance bar for an auction catalog is set by the worst moment it has to work in, not the average one.

The targets worth holding yourself to

Google’s Core Web Vitals give three public thresholds that translate directly into catalog design decisions: a good Largest Contentful Paint is under 2.5 seconds, a good Interaction to Next Paint is under 200 milliseconds, and a good Cumulative Layout Shift is under 0.1. For a catalog grid, LCP is almost always the first row of lot photos, INP is what happens when someone taps a filter, and CLS is the page jumping as images load in and push the lot they were reading off the screen.

Those three numbers describe most of what makes a catalog feel good or bad on a phone. Everything in the next section exists to hold them.

Photo weight is the whole game

A catalog page is mostly photographs, so the difference between a catalog that opens instantly and one that stalls is almost entirely how many bytes of image you send and when you send them. Nothing else on the page comes close.

Formats and sizes

Serve modern formats first. AVIF and WebP both compress substantially better than JPEG at the same visual quality, and web.dev’s image guidance covers the tradeoffs between them in detail. The practical setup is AVIF with a WebP or JPEG fallback, generated automatically when a photo is uploaded, so nobody on the sale team has to think about formats.

Generate several sizes of every photo, not one. A catalog grid on a phone needs an image a few hundred pixels wide. Sending the same four thousand pixel photo the consignor’s phone produced, then letting the browser shrink it, wastes most of the download on pixels nobody sees. Responsive image markup that offers the browser a set of widths lets each device take only what it needs.

As an illustrative target rather than a rule, a catalog grid thumbnail that lands somewhere around forty kilobytes and a full lot photo around a hundred and fifty will feel fast on mobile data. The exact numbers matter less than having a target at all, because without one the photos get bigger every sale until someone notices.

How many photos per lot, and where they load

Three to six photos suits most lots, more for anything where condition drives the bid. A serial plate, an hour meter, a maker’s mark, the damage: those are the photos that decide whether someone places an absentee bid or stays home.

The rule that actually protects performance is about placement, not count. The grid shows one thumbnail per lot. Everything else loads on the lot page, or in a gallery that fetches images only when opened. A catalog grid that eagerly loads a six-photo gallery per lot for a two hundred lot estate sale is sending more than a thousand images to a phone in a farmyard, and that is how a catalog becomes unusable at exactly the wrong moment.

Lazy loading, and what must not be lazy

Use the browser’s built-in lazy loading on every image below the first screen. It is a single attribute, it needs no script, and it means a bidder scrolling to lot 180 has downloaded lots 1 to 180 and nothing beyond.

The important exception: do not lazy load the images in the first visible row. Those are your Largest Contentful Paint. Marking them lazy delays the very thing the page is measured on. The first row loads eagerly, everything after it waits.

Reserving space so the page never jumps

Give every image explicit width and height attributes, or a CSS aspect ratio, so the browser reserves the space before the photo arrives. Without it, each image that loads shoves the content below it down the page. On a catalog that means a bidder reading lot 46 gets thrown to lot 52 mid-sentence, and it is the single most irritating thing a slow catalog does. It is also directly what Cumulative Layout Shift measures.

Alt text is not optional, and it is not decoration

Every lot photo needs a real text alternative describing the lot, not the filename the camera produced. The W3C’s images tutorial sets out how to write one that carries the same information the image does. For a catalog this pays twice: it makes the page usable for a bidder with a screen reader, and it gives search engines the description they need, which Google’s image guidance treats as one of the main signals for understanding what a picture shows.

Grouping and filtering that match how the sale is run

Group lots the way the sale is actually organised, because that is the order the bidder will experience them in. A catalog whose structure disagrees with the ring order is a catalog the bidder has to translate in their head all day.

Group by ring order first

Most sales run in a fixed order, and the lot numbers follow it. Whatever else you offer, the default view of the catalog should be lot number ascending, because that is the order the auctioneer will call and the order the printed catalogue will show. Clever default sorting, by price estimate or by popularity, makes a catalog worse for the person standing in the ring.

Where a sale runs multiple rings at once, which is common on large farm and heavy equipment sales and government surplus sales, the ring belongs in the grouping. A bidder who only cares about the shop equipment ring should be able to see that ring and nothing else.

The filters worth building

Filters earn their place when they match a decision a bidder is actually making. In practice that is a short list, and it differs by sale type:

Sale type	Filters that matter
Estate sale	Category, room or collection, lot number range
Farm and heavy equipment	Equipment class, manufacturer, year, hours or condition band
Fine art and antiques	Maker or artist, medium, period, provenance present
Government surplus	Asset class, department, location, condition code
Industrial liquidation	Machine type, manufacturer, plant or line, removal group
Benefit auction	Package type, experience or item, live or silent

Two filters most catalogs should also carry regardless of sale type: lots with photos, and lots still open in a timed sale. Both answer a real question quickly.

Keep filters instant. If tapping a filter means a full page reload over a weak connection, bidders will stop using them after the second try. Filtering that happens in the browser on data already loaded is the difference between a tool and an obstacle.

Search, and what it needs to match

Catalog search has to match on lot number first, then title, then description. A bidder who types “214” wants lot 214, not every lot whose description mentions 214 hours. Getting that precedence right takes ten minutes to build and saves a bidder twenty seconds every time they use it, which over a preview hall full of people is the difference between a catalog they trust and one they abandon for the printed sheet.

Per-lot pages versus one long list

Give a lot its own page when someone might search for that lot by name, and leave the rest as rows in the list. This is a judgment call per sale, not a setting, and both extremes are wrong.

The lots that earn a page

A per-lot page is worth building for anything with a name, a maker, a model or a story. A tractor with a model number. A vehicle with a VIN. A signed piece. A named collection. A piece of production machinery that someone in the industry searches for by model. These are lots where a stranger might find your sale through a search for the thing itself rather than a search for an auction, and that stranger is worth having.

A per-lot page also gives you somewhere to put the full condition report, the complete photo gallery, the provenance, the reserve status and a direct link to bid on that lot. It is the page you paste into a message when a buyer asks about one specific item.

The lots that do not

Box lots, shop miscellany, pallets of assorted stock and the long tail of a house clearance do not need pages. Nobody searches for “box of assorted hand tools”, and building three hundred near-identical thin pages makes the catalog slower to browse and gives a search engine three hundred pages with almost nothing on them.

The practical split on a typical estate sale is perhaps a couple of dozen lots with pages and the rest as rows. On a farm and heavy equipment sale it may be most of the catalog, because most lots are titled equipment. On an industrial liquidation it is usually the machinery and not the tooling.

The rule that keeps this maintainable

Whichever split you choose, the row and the page must come from the same lot record. If the grid row and the lot page are edited separately, they will disagree within two sales, and a description that disagrees with itself is worse than a short one.

Where condition reports and provenance belong

Condition reports belong directly under the title and photos on the lot page, in plain visible text, not behind a link and not in a downloadable file. A bidder deciding whether to place an absentee bid or drive two hours to a preview reads the condition report before anything else on the page, and every extra tap between them and it costs you bidders.

Write condition as facts rather than adjectives. “Runs and drives, hour meter shows the reading in the photo, hydraulic leak at the left rear cylinder” tells a bidder something. “Good condition” does not, and it invites a dispute after the hammer. What your house is willing to state about condition is your call and your liability, not a website decision, and nothing here is bidding, appraisal or valuation advice.

Provenance sits in the same block for lots where it moves the bid, written as text on the page rather than as an image of a document. If you have a document, publish both: the readable text and a link to the scan. Text is what a search engine can read and what a bidder can search within.

Two more things belong in that block and are often missing. Reserve status, stated plainly as subject to reserve or selling absolute, because it changes whether a bidder bothers. And the buyer’s premium, repeated on the lot page, because a bidder doing arithmetic on a hammer price should not have to open a second page to finish the sum.

Writing a lot title bidders actually search for

Write the lot title the way a buyer would describe the thing to someone else, front-loading the words they would type. The pattern that works across nearly every sale type is: year, maker, model, item type, then the one detail that distinguishes it.

An illustrative pair shows the difference. “Lot 47: Tractor” tells a search engine nothing and a bidder almost nothing. “Lot 47: John Deere 4010 diesel tractor, wide front, hour meter reading in photos” is searchable, scannable and honest about what is known.

Three rules keep titles useful:

1. Put the specific words first. Maker and model before generic category. A phone truncates a long title, and the truncation always lands at the end.
2. Do not stuff keywords. A title reading “tractor farm equipment auction tractor for sale” helps nobody and reads as spam to both a bidder and a crawler.
3. Keep the lot number in the title. It is the number the auctioneer calls and the number the bidder writes on their paddle card, and having it in the title means it is in the page title, the browser tab and the search result.

Where a lot has a genuine identifier, a model number, a VIN or a serial, put it in the description in full. Those are the highest-intent searches a catalog ever receives, because the person typing a model number knows exactly what they want.

Pagination or infinite scroll

Use pagination. Infinite scroll suits feeds where a reader is browsing without a destination, and Nielsen Norman Group’s analysis of infinite scrolling points out the same tradeoff that makes it wrong here: it works against goal-oriented users who need to locate a specific item and return to it later. That is exactly a bidder with a lot number.

Pagination gives three things a catalog needs. Each page has a stable URL, so a bidder can bookmark lots 200 to 250 and come back after the preview. Memory works, because “it was near the bottom of page four” is a thing a person can hold. And the footer stays reachable, which matters because the terms, the removal deadline and the contact number live there.

The compromise worth considering is a paginated catalog with a generous page size, around fifty to a hundred lots per page depending on how heavy your rows are, plus a “load more” button rather than automatic loading. The button keeps the user in control and keeps the page from growing without limit on a phone with little memory.

Whatever you choose, keep a jump-to-lot control at the top. The fastest path from a bidder to lot 214 is a box they type 214 into.

The printable catalogue for the preview hall

Generate a printable catalogue from the same lot data as the website, because in the preview hall paper still wins. People walk a yard or a hall with a printed list, mark it up, circle lots, write their own maximums next to them, and bid off those notes. Signal inside a steel building or a barn is often unusable, and a printed sheet needs none.

The printable version should be produced from the lot records automatically, not maintained separately. That single decision is what stops the printed catalogue from ever disagreeing with the site, which is the failure mode that causes arguments at the clerk’s table.

A workable printed catalogue is plainer than the web version:

• Lot number, title and a one-line description, set tightly so a page holds a useful number of lots.
• A small photo only for the lots that need one. Photos are what make a printed catalogue expensive and heavy.
• The sale header repeated on every page: house name, sale date, location, start time and the buyer’s premium.
• Blank space in the right margin. Bidders write in it, and a catalogue with nowhere to write gets replaced by the bidder’s own notebook.
• Terms and removal deadline on the last page.

Offer it as a print-friendly page as well as a downloadable file. Someone printing from a phone at home should get a clean result without needing to open anything else, and a print stylesheet on the catalog page is a small piece of work that does that.

A weak-signal test you can run this week

Take your catalog somewhere with bad signal, ideally the kind of building you actually hold previews in, and run these five checks on a phone.

1. Time the first screen of lots. If nothing is visible after a few seconds, the grid is sending too many bytes before it shows anything.
2. Scroll to lot 150 and back. If the page jumps while images load, image dimensions are not reserved.
3. Tap a filter. If it triggers a full reload, filtering is happening on the server and will be unusable in the field.
4. Try to reach one specific lot by number. If the only route is scrolling, add a jump-to-lot control and pagination.
5. Open a lot and look for the condition report without scrolling. If it is below the fold or behind a link, move it up.

Most catalogs fail the first and the fourth. Both are fixable without redesigning anything a visitor sees.

Where to take this next

If your catalog already holds up in a barn with two bars, there is nothing here to act on. If it does not, the work is usually image handling and page structure rather than a new look, and what is included on each plan is listed on the pricing page, with the one-time build fee invoiced only after the site is live.

If you would rather see a catalog built for your sale type than read about one, book a demo and we will open a working example on your own phone, on your own connection. We build and manage the website. The lots, the descriptions, the reserves and the ring stay entirely yours.